Study on Site Specific Strong Ground Motion for Seismic Retrofit Design

A STUDY ON SITE-SPECIFIC STRONG GROUND MOTION FOR SEISMIC RETROFIT DESIGN OF THE HANSHIN EXPRESSWAY LONG-SPAN BRIDGES IN OSAKA BAY AREA

Tsutomu NISHIOKA*, Toshihiko NAGANUMA*, Hidesada KANAJI* and Takao KAGAWA**

ABSTRACT The site-specific strong ground motion in Osaka Bay Area is studied in order to estimate the seismic load for the seismic retrofit of the Hanshin Expressway long-span bridges. We consider the crustal earthquake and the plate boundary earthquake as the maximum credible earthquake deterministically. The empirical Green’s function method and the stochastic semi-empirical Green’s function method are used to predict the site-specific ground motion. The study shows that there are some differences in the acceleration response spectra between the predicted strong ground motion and the standard ground motion in the Japanese specifications for highway bridges.

1. INTRODUCTION Many bridges suffered heavy damage from the Hyogoken-nanbu earthquake that struck the Kobe city in Japan on Jan. 17, 1995. The seismic retrofit program for the existing bridges made a full-fledged start slightly behind the restoration of the damaged structures in the wake of the disaster in Hanshin Expressway Public Corporation. Until today, we have already strengthened most of the middle- or small-scale brides seismically. But the long-span bridges, mainly located in Bay Route of the Hanshin Expressway network, have not retrofitted yet due to the financial problems. Although most of the highway bridges are retrofitted seismically according to the standard seismic loads in the current Japanese specifications for highway bridges [JRA(2002)], it is desirable in view of earthquake engineering that the input ground motion for the seismic design of a structure is estimated in consideration of source, path and local site effects, shown in Figure 1. In this paper, the site specific strong ground motion for seismic retrofit design of the Hanshin Expressway long-span bridges are estimated and compared with the Japanese specifications for highway bridges in terms of the acceleration response spectra.

2. THE LONG-SPAN BRIDGES IN HANSHIN EXPRESSWAY NETWORK The 18 sites of the Hanshin Expressway long-span bridges, mainly located along the coastline of Osaka Bay, are the objective sites to predict the strong ground motion. Table 1 is the list of the 18 long-span bridges, indicating regions, bridge names, bridge types, abbreviations of each bridge and longer natural periods between the longitudinal and transverse direction. The locations of the 18 bridges are shown in Figure 2.

* Hanshin Expressway Public Corporation ** Geo-Research Institute

Attenuated Amplified ground ground motion motion Near-source ground motion

Fault rupture Local site effects n atio ag s op ect Pr eff th pa Source effects

Figure 1 Schematic diagram of source, path and local site effects

Table 1 List of the 18 Hanshin Expressway long-span bridges

Region Bridge Type Abbrev. Period (s) Kobe Rokko Island Lohse arch rar 2.5 Higashikobe Cable stayed hkb 4.9 Shinashiyagawa Girder sas 2.5 Nishinomiyako Nielsen arch nmk 2.0 North Amagasakiko Girder amk 1.8 Osaka Nakajimagawa Nielsen arch njk 1.2 Kanzakigawa Nielsen arch kzk 1.2 Shorenji Girder srj 1.8 Central Umemachi V-shape uto 1.6 Osaka rigid frame Tempozan Cable stayed tbz 3.8 Minato Gerber truss mnt 4.6 Nankosuiro Lohse arch msr 1.4 South Hirabayashi Girder hrb 2.6 Osaka Yamatogawa1 Cable stayed ymt 3.1 Yamatogawa2 PC Box yms 2.5 Girder Shinhamadera Nielsen arch sht 1.9 Far South Harukigawa Girder hkk 2.5 Osaka Kisiwada Nielsen arch kwt 2.9

Figure 2 Locations of the 18 Hanshin Expressway long-span bridges

3. ACTIVE FAULTS IN OSAKA BAY AREA AND PLATE BOUNDARY FAULTS There are 6 active faults recognized in Osaka Bay area, shown in Figure 3. As Rokko-Awaji faults triggered the 1995 Hogoken-nanbu earthquake only 8 years ago, they are neglected in this study. Out of the other 5 faults, Uemachi faults and Osaka Bay faults are selected on the basis of the peak ground acceleration (PGA) attenuation study. In addition to the active faults in Osaka Bay Area, we consider the plate boundary faults located off the southwestern coast of Japan Islands. We adopt the fault model studied by the central government to predict the strong ground motion of the plate boundary earthquake. The fault model is shown in Figure 4.

Arima-Takatsuki Techtonic Line

s t l

s a Rokko-Awaji Faults lt Ikoma Faults u F a i F h c

y a a B m - e a k U a s O

Median Techtonic Line

Figure 3 Active faults recognized in Osaka Bay Area

Epicenter

Figure 4 Fault model of the plate boundary earthquake

4 4. SYNTHESIS PROCEDURE OF STRONG GROUND MOTIONS The strong ground motions are estimated for the crustal earthquake and the plate boundary earthquake respectively. We suppose that the active faults should cause the crustal earthquake and that the plate boundary faults should bring about the earthquake due to the continental plate movement. Two methods are used for simulating strong ground motion at a specific site resulted from a large earthquake. One is the empirical Green’s function method [Irikura(1986)] and the other is the stochastic semi-empirical Green’s function method [Boore(1983), Kamae and Irikura(1992)]. The empirical Green’s function method is one of the most effective techniques for predicting strong ground motion for the purpose of seismic design of a structure. This method synthesizes ground motion at a site from observed records as empirical Green’s function. A large earthquake is constructed by a superposition of many small earthquakes based on the regional scaling relation of source parameters. Figure 5 shows the schematic source model for synthesis. Since the method considers propagation path effects, local site effects, rupture propagation effects on the fault plane and geometrical relation effects between source and site, it is very powerful for predicting the strong ground motion by a future large earthquake. But appropriate records in the objective site are not always available as the empirical Green’s function. The stochastic semi-empirical Green’s function method is a revised simulation method when no suitable observation data are available as empirical Green’s function. This method uses the stochastically simulated small-event motion based on seismological spectral model as semi-empirical Green’s function, instead of the actual observed records. Since appropriate observation records don’t exist as empirical Green’s functions for the crustal earthquakes generated by Uemachi faults and Osaka Bay faults, we use the stochastic semi-empirical Green’s function method to predict the ground motion of the crustal earthquake. On the other hand, we obtained suitable records in the objective region for the plate boundary earthquake. We use the empirical Green’s function method to predict the ground motion of the plate boundary earthquake. For the crustal earthquake, we consider 3-D topography of deep geologic structure in Osaka Bay Area and calculate the ground motion by the 3-D elastic finite-difference model [Graves(1996), Pitarka(1999)]. Figure 6 shows the 3-D topography of deep geologic structure in Osaka Bay Area. We simulate the ground motion of the crustal earthquake in use of the hybrid method. That is to combine the short period range (0.1-2.0sec.) by the stochastic semi-empirical Green’s function method and the long period range (1.0-10.0sec.) by the 3-D elastic finite-difference model. This hybrid method takes the advantages of the two methods. The prediction procedure in this study is shown in Figure 7.

Figure 5 Schematic source model for synthesis

unit : m

Figure 6 3-D topography of deep geologic structure in Osaka Bay Area

Crsutal earthquake Plate boundary earthquake

Active faults in 6 active faults Source parameters studied by Osaka Bay area recognized Central Government

Attenuation PGA relationship

Select 2 faults Uemachi faults that have Osaka Bay faults larger P.G.A. Empirical Green's function method Semi-empirical 3-D finite-difference Green's function model method in short period range in long period range (0.1-2.0 sec.) (1.0-10 sec.)

Strong ground motions The hybrid method for seismic retrofit

Figure 7 Prediction procedure

5. STRONG GROUND MOTIONS OF THE CRUSTAL EARTHQUAKE The source parameters of Uemachi faults and Osaka Bay faults are shown in Table 2. Uemachi faults consist of 5 segments. The moment magnitude of the maximum credible earthquake (MCE) by Uemachi faults is assumed to be Mw=6.9. Osaka Bay faults are made up of 2 segments and the moment magnitude of MCE is Mw=6.7. For synthesizing the strong ground motion, we assume 5 fault rupture scenarios to Uemachi faults and Osaka Bay faults. The 5 types of the asperities are shown in Figure 8. These assumptions are based on Irikura(2002)’s research. The acceleration response spectra SA of the synthesized ground motion are calculated in the 5 regions (Kobe, North Osaka, Central Osaka, South Osaka and Far south Osaka). The non-linearity of the seismic response of the surface ground is taken into account by the SHAKE program. Figures 9-13 are the SA of the simulated ground motion. The spectrum of the Japanese specifications for highway bridges, drawn with broken line in Figures 9-13, is the one of the strong ground motion of the maximum credible crustal earthquake in the soil classification 3 [JRA(2002)]. The comparison shows that most of the SA of the site-specific ground motion in the short period range are smaller than those of the

7 specification. However, in some longer periods, the SA of the site-specific ground motion go beyond those of the specification. Since the long-span bridges have long natural periods, we should pay much attention to the SA of the site-specific ground motion in the long period range when we conduct the seismic retrofit design of those structures.

Table 2 Source parameters of Uemachi faults and Osaka Bay faults Osaka Bay Uemachi faults faults Part A Part B Part C Part D Part E Part A Part B Length (km) 8 12 4 6 24 26 10 Width (km) 22 16 Strike (deg.) -16.2 7.4 48.4 55.8 21.4 -158 -114 Dip (deg.) 60 80 Moment 6.9 6.7 magnitude Mw Seismic moment 0.38 0.58 1.29 0.32 0.19 1.03 0.39 (1019 N･m)

(a) Uemachi faults, Type 1

(b) Uemachi faults, Type2

Figure 8 Assumed asperities in Uemachi faults and Osaka Bay faults

Part A Part B Part C Part D Part E

Part A Part B

(e) Osaka Bay faults, Type5

Figure 8 Assumed asperities in Uemachi faults and Osaka Bay faults